69Dividend Policy under Asymmetric Information

DividendPolicyunderAsymmetricInformationMertonH.Miller;KevinRockTheJournalofFinance,Vol.40,No.4.(Sep.,1985),pp.1031-1051.StableURL:=0022-1082%28198509%2940%3A4%3C1031%3ADPUAI%3E2.0.CO%3B2-4TheJournalofFinanceiscurrentlypublishedbyAmericanFinanceAssociation.YouruseoftheJSTORarchiveindicatesyouracceptanceofJSTOR'sTermsandConditionsofUse,availableat://@jstor.org.:152007THEJOURNALOFFINANCEVOL.XL,NO.4SEPTEMBER1985DividendPolicyunderAsymmetricInformationMERTONH.MILLERandKEVINROCK*ABSTRACTWeextendthestandardfinancemodelofthefirm'sdividend/investment/financingdecisionsbyallowingthefirm'smanagerstoknowmorethanoutsideinvestorsaboutthetruestateofthefirm'scurrentearnings.Theextensionendogenizesthedividend(andfinancing)announcementeffectsamplydocumentedinrecentresearch.Butoncetradingofsharesisadmittedtothemodelalongwithasymmetricinformation,thefamiliarFisheriancriterionforoptimalinvestmentbecomestimeinconsistent:themarket'sbeliefthatthefirmisfollowingtheFisherrulecreatesincentivestoviolatetherule.Weshowthataninformationallyconsistentsignallingequilibriumexistsunderasymmetricinformationandthetradingofsharesthatrestoresthetimeconsistencyofinvestmentpolicy,butleadsingeneraltolowerlevelsofinvestmentthantheoptimumachievableunderfullinformationand/ornotrading.Contractualprovisionsthatchangetheinformationasymmetryorthepossibilityofprofitingfromitcouldeliminateboththetimeinconsistencyandtheinefficiencyininvestmentpolicies,butthesecontractualprovisionstooarelikelytoinvolvedead-weightcosts.Establishingwhichrouteorcombinationofroutesservesinpracticetomaintainconsistencyremainsforfutureresearch.THESTANDARDFINANCEMODELofoptimalinvestment/financing/dividendde-cisionsforthefirm(assummarized,say,inFamaandMiller[16],Chapters2and3whichinturnbuildsontheearlierworkofMillerandModigliani[33])assumes,amongotherthings,thatoutsideinvestorsandinsidemanagershavethesameinformationaboutthefirm'scurrentearningsandfutureopportunities.WeproposeinSectionIofthispapertoreplacethatassumptionwiththemoreplausibleonethatmanagersknowmorethanoutsideinvestorsaboutthetruestateofthefirm'scurrentearnings.Forthetheoryoffinance,thatreplacementbringsbothgoodnewsandbadnews.Thegoodnewsisthatdividend(andfinancing)announcementeffects,amplydocumentedinrecentempiricalresearch,nowbecomeimplicationsofthebasicdecisionmodelratherthanqualificationsappendedtoitasintheoriginalMiller-Modigliani(hereafterreferredtoasMM)treatment[33,p.4301.Inaworldofrationalexpectations,thefirm'sdividend(orfinancing)announcementsprovidejustenoughpiecesofthefirm'ssourcesandusesstatementforthemarkettodeducetheunobservedpiece,towit,thefirm'scurrentearnings.Themarket'sestimateofcurrentearningscontributesinturntotheestimateoftheexpected*UniversityofChicagoandHarvardUniversity,respectively.WeacknowledgewiththankshelpfulcommentsonearlierversionsbyPaulAsquith,MichaelBrennan,SudiptoBhattacharya,GeorgeConstantinides,DouglasDiamond,ChrisostomoGarcia,JohnGould,SanfordGrossman,BengtHolmstrom,GurHuberman,JonathanIngersoll,KoseJohn,CharlesKahn,EdwardLazear,JamesOhlson,ArturRaviv,RichardRoll,RobertVerrecchia,JeroldWarner,andJosephWilliams.10311032TheJournalofFinancefutureearningsonwhichthefirm'smarketvaluelargelyhinges.Themodel'sdividendinformationeffectsarethusentirelyconsistentbothwiththeMMpropositionthatthevalueofthefirmisgovernedbyitsearningsandearningpower;aswellaswiththefindingsofWatts[44]andGonedes[17]thatintime-seriesforecastsoffutureearnings,currentandpastdividendsappeartohavelittlepredictivepoweroverandabovecurrentandpastearnings.ThebadnewsisthatthepriceofallowingforinformationasymmetryanddividendannouncementeffectsmaybethelossofthefamiliarFisheriancriterionforoptimalinvestmentbythefirm-viz.,investinrealassetsuntilthemarginalinternalrateofreturnequalstheappropriatelyrisk-adjustedrateofreturnonsecurities.Inaworldwherethemarkettakesannounceddividends(orfinancing)asacluetounobservedearnings,temptationsarisetorunupthepricebypayingoutmoredividends(orengaginginlessoutsidefinancing)thanthemarketwasexpecting,evenifthatmeanscuttingba